eta - eta' - glueball mixing

FTUV 10-0301

T] — T]'- glueball mixing

Simon Kiesewetter and Vicente Vento

Departament de Fisica Teorica and Institut de Fisica Corpuscular,
Universitat de Valencia-CSIC, E-46100 Burjassot (Valencia), Spain.

Abstract

We have revisited glueball mixing with the pseudoscalar mesons in the MIT bag model
scheme. The calculation has been performed in the spherical cavity approximation to the
bag using two different fermion propagators, the cavity and the free propagators. We obtain
a very small probability of mixing for the rj at the level of 0.04 — 0.1% and a bigger for the rj'
at the level of 4 — 12%. Our results differ from previous calculations in the same scheme but
seem to agree with the experimental analysis. We discuss the origin of our difference which
stems from the treatment of our time integrations.

Pacs: 12.39.Mk,12.39.Ba,14.40.-n,14.65.Bt

Keywords: Glueball, Bag Model, Mesons, Mixing, Quarks

Email: name.surname@uv.es

1 Introduction

Quantum Chromodynamics (QCD) is the theory of the hadronic interactions. It is an elegant
theory whose fuh non perturbative solution has escaped our knowledge since its formulation more
than 30 years ago [1]. The theory is asymptotically free [2|13J and confining A particularly
good test of our understanding of the nonperturbative aspects of QCD is to study particles where
the gauge field plays a more important dynamical role than in the standard hadrons. For this
reason the glueball spectrum has attracted much attention [5J. The interest in this subject is
related to the significant progress in the understanding of the properties of such states within
QCD, as well as, in the new possibilities for their identification in modern experiments.

From the phenomenological point of view it has become clear by now that it is difficult to
single out which states of the hadronic spectrum are glueballs because we lack the necessary
knowledge to determine their decay properties. Moreover the strong expected mixing between
glueballs and quark states leads to a broadening of the possible glueball states which does not
simplify their isolation. The wishful sharp resonances which would confer the glueball spectra
the beauty and richness of the baryonic and mesonic spectra are lacking. This confusing picture
has led to a loss of theoretical and experimental interest in these hadronic states. However, it is
important to stress, that if they were to exist they would be a beautiful and unique consequence of
QCD. A comprehensive review on the experimental status of glueballs has recently appeared [6j.

For the purposes of this paper we accept the existence of at least one pseudoscalar glueball
state although its existence has been a matter of debate since the Mark II experiment proposed
glueball candidates [7]. Note that the pseudoscalar sector is a complex one. On the one hand
it accommodates the Goldstone nature of the pseudoscalar multiplet, on the other, not totally
unrelated, we encounter the singlet-octet mixing, which is traditionally associated with the res-
olution of C/(l) anomaly. In constituent models the ideal mixing {6i = tan~^ \/2) is natural,
however the r] and rj' mixing is non ideal.

Gluon self-couplings in QCD suggest the existence of glueballs, bound states of mainly glu-
ons [1]. Investigating glueball physics requires an intimate knowledge of the confining QCD
vacuum and it is well known that such properties cannot be obtained using standard perturba-
tive techniques. To handle the non-perturbative regime of QCD, one can resort to numerical
methods, known as lattice QCD. Lattice QCD needs as input the quark masses and an overall
scale, conventionally given by Aqcd- Then any Green function can be evaluated by taking aver-
age of suitable combinations of lattice fields in vacuum samples. This allows masses and matrix
elements, particularly those of weak or electromagnetic currents, to be studied. However lattice
QCD faces both computational and fundamental problems in the description of glueballs [5j.
A complementary way to describe glueballs, namely the MIT bag model, implements in a dy-
namical way the phenomenological properties of the confining QCD vacuum and the interaction
among the gluons. Historically the investigation of the glueball properties started precisely in
this model Jaffe and Johnson found many glueball states with different quantum numbers
lying in the mass interval 1000-2000 MeV. They emphasized that one should expect rather small
widths for such states because their decays in conventional hadrons violate the Okubo-Zweig-
lizuka (OZI) rule |10].

The aim of the present investigation is the study of the mixing between a possible pseudoscalar
glueball state and the r] or r/'-mesons. The calculation has been performed in the MIT bag model,
a description which imposes by fiat some of the properties of QCD. In this model a hadron is
basically a bubble of perturbative vacuum in the midst of a non-perturbative vacuum. Inside

1

the bubble we insert the constituents, which are described by cavity modes, and the surface of
the bubble screens color from flowing into the non-perturbative world. The calculation has been
performed in the so-called spherical cavity approximation, where several improvements have been
incorporated, like center-of-mass corrections and the recoil correction. In this setup, the cavity
is fixed to be a sphere and its radius is allowed to vary dynamically. Within this scheme we have
performed two calculations of the mixing Hamiltonian. One, in which we have used the cavity
propagator for the quarks . This cavity propagator is made up of a sum over all possible cavity
states. Thus it incorporates, in principle, the confining property of the bag model. Another, in
which we have used the free propagator which is made up as a sum over free modes. As it turns
out, the results of both calculations are almost the same, so the dominating property, at least
for the problem investigated here, is asymptotic freedom.

Our investigation is presented as follows. In [2] we show the necessary tools to carry out
the calculation. Starting from the QCD Lagrangian we use a formalism that allows one to
calculate the mixing energies perturbatively by means of the appropriate Feynman diagrams.
We have to introduce for this purpose the bare glueball and meson states. In sec. 12. 3| we discuss
the quantization, which is important since the problem at hands is a multi-particle one. After
discussing the role of the propagator in sec. 12. 3^ and addressing and resolving an important
physical problem that arises in bag model calculations in sec. 12.41 we present and comment on
the results in sec. [3] and give some conclusions in sec. 3] The actual calculations have been
relegated to the appendix to ease the reading of the main text.

2 Calculation of the Ty-zy'-glueball mixing in the bag model

We next calculate the mixing energy, which corresponds to off-diagonal Hamiltonian terms in
Fock space. In subsection 12.11 we introduce a formalism which allows the calculation of the
mixing energies in a perturbative manner. In subsection 12.21 we present the bare glueball and
meson states in the bag model. Thereafter we discuss important aspects of the calculation

2.1 Formalism

QCD is a non-abelian Yang-Mills theory with a SU{3) gauge symmetry regarding color charge.
The Lagrangian is

£QCD = i^{iIp-m)^l;-^F^,Fr

(1)

where

d^-igAl.t-

(2)
(3)

2

= 2 A" where A'^ are the Gell-Mann matrices and f are the structure constants of the SU{3)
algebra. Some rearrangement yields

1

QCD

where we can identify the free Lagrangian of QED (Cq) with color indices, a quark-quark-gluon
vertex a 3-gluon vertex (£2) and a 4-gluon vertex (£3).

We use a perturbative approach inside the cavity following the scheme developed by Maxwell
and Vento The glueball and meson states represent solutions of the free Lagrangian Cq^

and > respectively. The equations of motion arising from the Lagrangian of QCD are

and

(5)

(6)

Eqs.(l5|) and ^ can be understood as an inhomogenous Dirac equation and Maxwell equation,
respectively. Thus, they can be solved exactly using the Feynman propagator for the Dirac field
and the Maxwell field, respectively in the following way

^(x) = g / d^x'SF{x,x')j^'Al{x')ei^{x')

and

(7)

(8)

A^^ix) = g J d^x'DF{x,x') [-r^'^d^ {A'i{x')A-,{x')) - r''Al{x')Fr{x') - ^Pi^t^^
One can now expand ^{x) and A^^{x) in a power series of g and obtains for the first order term

= +g f d'^x'SF{x,x')-f^'Af^{x')ei}^'^\x') (9)

= +g y d'x'DF{x,x') [-r'^d, [Af^{x')Af>{x')

(10)

One can now use the expressions ^^^"^ and A^^^ or higher orders of and A to calculate the
expectation value ^ of some observable f to various perturbative orders of g. In here we are

interested in the expectation value of the quark-gluon interaction Hamiltonian Hi = gip^^^ A'^t""ip .

3

-AAAAAAAAAA/''" "

-Vwvwvvw'" '

Figure 1: Fermion exchange contribution to the meson-glueball mixing (left) and Gluon exchange
to lowest order gives no contribution(right).

Inserting i/j^^") yields

+g j d^xi^^^\x)-i^'A^^\xY^t''i)^^\x) (11)
= 2g^ [ d^xlp{x)-f^'A{x)X f d'^x'SF{x,x')j''A{x'ffy{x') (12)

with zero order wavefunctions in eq. (|12|) . This expression corresponds to an exchange of a
virtual fermion as shown in fig. [TJ Inserting different orders of ^ and A, one obtains expressions
corresponding to different Feynman diagrams. Every order in ip brings a fermion propagator,
while every order in A brings a gluon propagator. For our calculation, we will restrict ourselves to
the one-fermion exchange, since this is the leading-order diagram of the meson-gluon interactions.
The lowest-order gluon exchange diagram, shown in fig. [H does not contribute, because the gluon
is a spin-1 particle and therefore does not couple to the spin-0 pseudoscalar meson or glueball.

2.2 Glueball and meson states

In order to construct the glueball states, we have to describe the gluon cavity modes. The
4-vector potential is given by

= (^-A° = (t>,A^ (13)

Since we are working in the static cavity approximation, we will be using the Coulomb gauge

V-i'=0 (14)

In this gauge, the scalar potential is given by Poisson's equation

-A0=-^ (15)
eo

and thus vanishes since there are no free charges in the model considered here.

The solutions can be classified into two different classes, which are called transverse electric
(TE) and transverse magnetic (TM).

Furthermore, the solutions are classified by the quantum numbers

I = 1,2,... orbital excitation (16)
m= — 1,-|-1 magnetic quantum number (17)

4

The boundary conditions generate the constraints

= (18)

r=R

for the TE sohition and

3l{kR)=0 (19)

for the TM solutions with ji being a spherical bessel function of order /. The transcendental
equations (jlSp and (jl9p have an infinite number of solutions kR = Xn, labeled by the radial
quantum number n = 0, 1, .... The lowest modes of interest here are 1=1, n=0, xte = 2.74 and
XTAi = 4.49 [TT].

The parities of the modes are vr = ( — l)'"*"^ for the TE modes and vr = (—1)' for the TM
modes. The non-linear boundary condition requires / = 0. This, however, is incompatible with
the helicity of the gluon. Thus, we must assume that the best value for / is the lowest possible,
that is Z = 1. The lowest-lying pseudoscalar glueball with parity J'^'^ = ^, which is the
objective of our investigation, contains the lowest-lying TE mode and TM mode gluon.

Let ajf-^n denote the particle creation operator associated with the gluon cavity-state denoted
by the quantum numbers l,K,m,n, where k G {TE,TM} denotes the polarisation. The lowest-
lying state with k = TE is given by / = 1,?7- = and the lowest-lying state with k. = TM is
given by Z = l,n = 0. Thus, the glueball state can be constructed by

\G) = -j= (a^pEt^TMl ~ "■TEl^TMi) 1^) (^0)

where / = l,n = everywhere. We impose the restriction that the state be a color singlet.

The meson states are constructed from the cavity fermion modes. To find them, we have
to study the radial solutions of the free Dirac equation. They are characterized by their total
angular momentum j = 1/2,3/2,..., a magnetic quantum number m = —1/2,1/2 and another
quantum number A = —1, 1, called Dirac's quantum number. The wavefunctions are |12|

(iXjl (pr) \

v{x) = N ('^'' • "1 ] ^-(x)e'-* (22)

where I = J + and I' = J — ^A, Q = . In the case of a massive field, p and u are related

by u; = \/jP + rn?, otherwise p = to. Here u denotes the particle solution and v the antiparticle
solution. The object ^™ is a 2-spinor of total angular momentum j, projection m and orbital
angular momentum /, called spinor spherical harmonics, defined by

(x) = Yl ( ^^If'lJM > Yi^ (x) (23)

Inserting eq. (|2ip and eq. (j22p into the boundary condition yields the constraint

jHpR) = -^MpR) (24)

5

This transcendental equation has an infinite set of solutions for each combination of j, A. The non-
linear boundary condition requires j = \ for the quark states. The quark states are normalized
by the requirement

d^xv){x)u{x) = 1. (25)

V

This yields for the lowest-lying state (j = i, n = 0, A = — 1) a normalization constant of

2'

2 1 x{v — mR) 1

A^^ = ^ f26)

R^joixY 2x {D -l) + mRA-K ^ '

where x = ujR and u = pR. Note that this factor differs from the one used in |13) and thereafter
by many authors. We have checked numerically that this is the correct formula.

The //-mesons are pseudoscalar mesons, thus have parity J^'-" = '". Let b^xmn ^^'^ ^jXmn
denote the particle creation operator and anti-particle creation operator, respectively associated
with the fermion cavity-state denoted by the quantum numbers j, A, m, n. We are interested in
the lowest-lying meson state. The lowest energy modes are associated to the quantum numbers
j = = 0, X = —1. Because of intrinsic parity between particle and antiparticle states,
P = —1 can be obtained using the lowest-lying particle state and the lowest-lying antiparticle
state. Thus, the meson state can be constructed by

1

V2

\(t) = ^ {b\d\ - |0) (27)

where j = ^,n = 0, A = — 1 everywhere. We have used the notation f, \. for m = ^, — ^
Furthermore, we impose a color singlet state and a flavor composition of

ri = ^{uu + dd- 2ss) (28)
V6

r/' = (nn + dd + ss) . (29)

The above definition of rj and r]' is actually an approximation, since we know that due to the
chiral anomaly the physical rj and rj' states are mixtures with a small mixing angle of r]i and r/g.
We neglect this fact throughout this work and identify rj with r/g and rj' with rji, as done above.

2.3 Propagators

The off-diagonal term in the energy expectation value is given by

{G\N!.2g^ I d^xlp{x)j^'A{x)X I d'^x SF{x,x')-i^ A{x'tt''^{x')\ \a) (30)

where N denotes the normal ordering operator. This expression can be evaluated using eq.([2
and eq. (|20p when one quantizes eq. (jl2p in the manner

tp{x) tp{x) = ^ ^Uaix)ba + Va{x)di (31)
a

tp{x) ipix) = ^ |^?;„(x)da + Ua{x)b'l (32)

a

A{x) ^ A{x) = [A^ix)da + A*^{x)di\ (33)

6

with the commutation relations

Fermion:
Fermion:

Gluon: aa,a\

(34)

(35)

(36)

Evaluating eq. ([30|) with a normal-ordered operator and shifting h\ d) and a) to the left and 6, d
and d to the right yields

Fermion: d^^Sp^ - S^^dp^

(37)

Gluon: SiTE^S^TMi — SlTEI^kTMA; + ^LTMi^f^TE^ — ^LTMt^KTEi

(38)

where a refers to the sum index in tp with particle states, /3 to the sum index in ip with antiparticle
states, L to the sum index in the first gluon wavefunction A and k to the sum index in the second
gluon wavefunction A.

There are two possible choices for the propagator Sf{x,x') in eq. (|12p . namely the confined
propagator and the free propagator. The confined propagator is built of a complete set of
confined states in the bag model. The free propagator is the well-known Feymnan propagator
Sf{x, x') = J (27r^4 " ^^m+ie ^^^h the Fcyumau prescription of closing the contour. Both
propagators act as Green's functions with respect to the Dirac operator, thus formally both
propagators can be used. However, since they are of a very different shape, one should expect
them to generate different results. Physically, it is not clear which propagator is preferable. On
the one hand, virtual fermion are not constrained to the bag as the bound states are. On the
other hand, the bound states are not momentum eigenstates, so the problem does not resemble a
scattering process and there is no 4- momentum conservation at the vertices. These problematic
will be elaborated in more detailed in sec. I2.4l and with the recoil correction, a possible resolution
is presented. In a way, the confined propagator may overemphasize the aspect of confinement,
while the free propagator may overemphasize the aspect of asymptotic freedom in the bag model.
Out of curiosity, we will carry out the calculation using both propagators.

1. Confined propagator: Since the radial solutions of the Dirac equation eq. ()2ip and eq. (p2|)
form a complete set, we will follow an approach by Maxwell and Vento pjj and use them
to construct the confined propagator

with a = {n, A, j, m) denoting a multiindex. Note that since the solutions represent virtual
particles rather than real quark states, they are not subject to the non-linear boundary
condition and thus, values other than j = \ are possible.

a

(39)

7

fWVWVWW^-"

ujj) + A'

^AAAAAAAAA;

LOxE

Figure 2: Recoil correction; Energy A and A' has to be added to account for center-of-mass
motion that necessarily arises through emission and absorption of the virtual particle

2. Free propagator: The solutions generating the free propagator are not subject to the bound-
ary conditions. Dropping the boundary conditions yields

n,\,J,M A,J,M
Imposing the normalization condition

E ^ [dkYl (40)

xut{x)uki (x) = / d^xvk{x)vf^i{x) = 6{k — k') (41)

yields

u^,{x) = J -—^= . ) (x)e-*"* (42)

vu{x) = {'^'' ^^"^ {^-^^)\ (43)

where we have used

drr^ji{kr)ji{k'r) = - k') (44)

This way, the calculation for the free propagator is analogous to the one shown in the
appendix, wi«, ditoe... nonnaliza.i„„ of TM. conespond. directly

to Rayleigh's expansion of plane waves.
2.4 Time integration and recoil correction

As mentioned above, energy conservation on the vertices is not possible in the process described
here since 7^ ^te and 7^ oJtm- Thus, the integrations of eq. (|12p might be formally carried
out (although for specific quark masses, there arise unphysical divergences), but the physical
interpretation remains unclear. It is also unclear, why the dt'-integration (corresponding to the
lower vertex) has to be carried out but the dt-intergration doesn't, so that there will be an
energy denominator only related to the energies at the lower vertex. Apparently, there must be

8

a (physical) flaw somewhere. In fact, the flaw hes within the bag. When virtual particles that
violate energy conservation are being created, a center-of-mass motion must be expected that
ultimately alters the energies of the states. This process will be taken into account here and we
will call this the recoil correction. Through the emission and absorbtion of the virtual fermion
the states will obtain an additional center-of-mass motion energy A and A', respectively as seen
in fig. [21 so that we are in the center-of-energy frame. This gives

CJo + A -I- OJTM =

-uq — a' + ojte =

(45)
(46)

Only in the center-of-energy frame can the virtual fermion exchange be understood in a physically
plausible way. Making this approximation, my work differs from other works in the bag model,
where this issue has not been addressed.

Writing down the time dependence, which we have left out before and applying the recoil
correction yields

uo(x)e*('^«+^)*4TA/(2;)e-^""^*
-v^{x)v^{xy^-^'-'">e{t' -t)

dt'

^TEyx )e

u^{x)u^{x')e-'^"^'-''^e{t - t')
''"^^*'t;o(x')e'("o+^')*'

(47)

where we have left out the spatial integrations for convenience. Performing the dt' integration
yields

/ dt'e'^^^^'eit - t') = t dt'e'^"^' = —

J J-oo

dt'e-'^"^'eit' - t)

oo
oo

lOJr

dt'e-"^^^' = -- —

OO

t

-iUat

(48)
(49)

oja — *e implicitly. This gives an overall denominator of for each
mode which is being propagated

where we have shifted oj^

3 Results

After carrying out a detailed calculation, which can be found in the appendix, one obtains for
the mixing energy fig. [3] as a function of the quark mass times the bag radius. The shown
energy is per quark-pair, i.e. qq. To calculate the corresponding mixing energy one has to take
into account the wave function of the meson states. As mentioned already in sec. 12. 2| we are
neglecting the r/ — r/'-mixing and identify r] as r/g and rj' as rji. Please note that both calculations,
the one using the confined propagator, and the one using the free propagator, give very similar
results. From now on we will only use the confined propagator results.

The values of the glueball mass change dramatically in the literature from one calculation
to another. Lattice QCD in the quenched approximation leads to a value around mc = 2500
MeV |14ftl6j . Unquenched calculations should produce a lower value as happens for the scalar
glueball |17| . This has been shown to be the case in an effective theory calculation of glueball
mixing which reproduces a large amount of data [18], where the lower pseudoscalar mass value is
set at 2000 MeV. Other effective theory calculations which fit parameters to data lead to values
down to 1400 MeV [19].

9

Figure 3: Glueball- meson mixing energy per quark-antiquark pair as a function of mR for the
free and confined propagator

Kuti |20) suggested that a rehable glueball spectrum, which is in reasonable agreement with
lattice calculations, can be obtained for fiV* = 280 MeV and R ^ 0.5 fm = 2.5 GeV-^ He
gives a coupling constant as = 0.5 to obtains a ^ glueball mass of about 2500 MeV in sharp
contrast with the old calculation of Jaffe and Johnson [9j, who chose parameters closely related
to the baryon spectrum. In order to perform our mixing calculation we shall take for the mesons
the experimental values, r/(550) and t/'(960).

The mixing probability strongly depends on the glueball mass and it turns out to be an
analytic function of it.

1

1

256 N 2

1 1

where

2 2^1 +(^)^

r/'-G system
r]-G system

960 + A'
550 + A.

This comes about from the diagonalization of the matrices

960 128
128 960 + A'

550 -27
-27 550 + A

r]'-G system
r]-G system

(50)
(51)

(52)
(53)

(54)
(55)

10

when one puts in the angle Q = arctan ^ , 6 being the mixing energy, into the expression for the
mixing probabihty = sin(®).

In order to confront the experimental situation let us vary the mass of the glueball following
the bag model prescription, i.e.

AjuJE + UJM - Z)

^ = m ' ^'""^

where we have eliminated B by the pressure balance equation and we used the the lowest TE
and TM modes to calculate the glueball energy. The term Z represents the zero point energy,
which we fit to have a glueball mass of 2500 MeV at a radius of 0.5 fm |20| . We omit here
the perturbative contributions to the mass. This energy has been corrected for center of mass
spurious motion to obtain the particle mass which we show in fig. [3l Note that the calculation
connects the Kuti and Jaffe and Johnson value ranges for different values of the bag radius. We
show in the figure the mass of a light baryon calculated with the same zero point energy as
a function of radius and see that the it reaches 1100 MeV at i? = 1.0 fm, which is the right
value before perturbative OGE corrections for the Nucleon-Delta system. Thus we have found
a consistent approximate formula to zeroth order which ascribes the value of the glueball mass
to its size and which contains all of the results obtained by the different calulcations mentioned
above.

R(fm)

Figure 4: Variation of glueball and baryon masses with bag radius.

By using Eq. ([3]) we obtain the mixing energies shown in fig. |3l From these mixing energies
we can calculate the mixing probability by diagonalizing the mixing Hamiltonian, using for the
glueball mass Eq. [56l and the experimental rj and r/' masses. We show them in fig. [3l We have
assumed for the calculation of the mixing energies a constant as = 0.5. This is not the required
value to reproduce the baryons and mesons masses at around R = 1 fm. The value used in
these calculations is closer to 2.0 [9]. If the glueballs would behave in a similar manner for larger

11

radius, since the mixing energy is proportional to a^, it would increase by a factor of 4 and the
mixing probabilities by a factor of 2 — 3. Thus we are showing in the present calculation the
minimum values for the mixing probabilities.

0,5 0,6 0,7 0,8 0,9 1,0 0,5 0,6 0,7 0,8 0,9 1,0

R (fm) R (fm)

Figure 5: ?7-glueball and r/'-glueball mixing energies as a function of bag radius (left) and rj-
glueball and r/'-glueball probabilities as a function of bag radius (right)

It is worth mentioning, that our result deviates significantly from a prior analysis by Carlson
and Hansson |21) . The difference between our work and theirs is our use the recoil-correction as
in 12.41 In the figures there is an additional difference arising from the use of different parameters.

The results of fig. [3] are quite illuminating. In no case does the rj mix with the glueball, even
for large couplings and small glueball masses. On the contrary the rj' can mix up to 4% for small
glueball masses and up to 12% for large strong as- We obtain a scenario strongly dependent
on the glueball mass. If the glueball mass is close to the lattice value the pseudoscalar glueball
should arise as an almost pure state with very distinct features, if on the contrary the mass is
small it might mix with the rj' but never with the rj.

A caveat we have not considered in our calculation is the mixing between the r/g and rji as
mentioned before. This mixing would increase the mixing probability of ?7-glueball due to its ryi-
component and decrease the mixing of r/'-glueball, due to its r/g-component. Thus our calculation
point towards a scenario with small mixing of the pseudoscalar glueball with the pseudoscalar
mesons.

4 Conclusions

We have performed a calculation of the mixing of the pseudoscalar glueball with the pseudoscalar
mesons ry and rj' . Our work suggests that the mixing is small if the mass is around 2500 MeV.
In the framework of the bag model, this is a new result. A previous study by Carlson and
Hansson [21| suggests a much larger mixing. However, a small glueball mixing, in this glueball
mass range, is as well favored by recent studies |19[I22| . Accordingly, glueball mixing should
not play an important role in the rj-rj' mass splitting. This also implies that a rather undiluted,
well-defined, narrow and therefore long-living pseudoscalar glueball state should exist. If the

12

mass of the glueball is closer to that of the r/', around 1400 MeV, then the mixing is larger and
the consequences of phenomenological analyses should be reanalized, since the amount of mixing
would determine if the glueball behaves more like mesons and baryons, i.e. large objects.

The conceptual difference between our work and other bag model calculations, like the one
carried out by Carlson and Hansson |21) . is the recoil correction. The recoil correction is in our
opinion a necessary ingredient to carry out the calculations in a physically meaningful way by
avoiding spurious singularities. The bag model in the static spherical cavity approximation fails
in describing the creation and absorption of virtual particles, because in the spherical cavity, the
mode energies are discrete and fixed. Thus energy conservation at the vertices is generally not
possible. This subtle problem has been neglected in all previous bag model calculations so far.
In our calculation it manifested in an (unphysical) singularity dominating the results. Our way
to resolve this problems is to take into account the recoil that arises on particle emission and
absorption, manifesting itself necessarily in center-of-mass motion of the constituents. We add
this energy to the energy of the constituents. This not only eliminates the singularity and makes
way for meaningful results, but it is necessary to cancel out the time-dependence of the mixing
energy. Without the recoil correction, the mixing energy is oscillating, which is also a sign of
center-of-mass motion problems. All of these hints suggest that the recoil correction is correct,
physical meaningful and must be applied to all bag model calculations.

At this point it is worth mentioning a technical error which despite its simplicity has been
in the field for many years. The normalization constant eq. (l26p in the case of massive quarks
for the quark mode wavefunction was written incorrectly, most probably a typo, in the original
paper |13| . but the error has been carried on by all papers that we have used thereafter. Luckily
the error does not imply large effects in the calculations of the light quarks.

In our work, we have neglected the mixing between r] and rj' and rather identified the physical
states with the singlet and octet states. A more elaborated calculation would be a three-particle
mixing scheme between rj, ry' and the glueball. Nevertheless, since the mixing between the mesons
and the glueball that we have obtained is so small, a three-particle mixing scheme will hardly
produce very different results and therefore we have decided to settle for the two-particle mixing
scheme.

This work has set the standards for future calculations within the MIT bag model scheme.
Any calculation dealing with the spectrum or mixings will have to follow the same procedure
and approximations. In particular an interesting phenomena which we are now revisiting is the
mixing of r/g and r/i to build the true physical r/- mesons.

Acknowledgements

We are grateful to V. Mathieu for interesting discussions. SK thanks the Departamento de Fisica
Teorica of Valencia for the hospitality and Prof. Kunz-Drolshagen for giving him the opportunity
to work there. This work was supported in part by HadronPhysics2, a FP7-Integrating Activities
and Infrastructure Program of the European Commission under Grant 227431, by the MICINN
(Spain) grant FPA2007-65748-C02-1 and by GVPrometeo2009/129. The diagrams have been
drawn using Jaxodraw |23| .

13

Appendix: Details of the calculation

Substituting eq. ([39|) into eq. (jl2p yields a color structure of the form

where -0 means the charge-conjugated i}). The gluon color singlet gives a factor of 6ab
the meson color singlet a factor of Sai /\/3- This leads to the expression

Tr (A'^A")

4-^/3- \/8

Summation over a yields an additional factor of 8, which finally gives a factor of

3-

Disregarding the time dependence, we show how to calculate the spatial integrals

/ d^x^{x)Jkvo,{x) / d^x'va{x)A{?^)^{^)

with the combinations

A\ = Ate^, A2 = At Ml
■ {Ai = Atei, A2 = Atm^)
Ai = At Mi, M = Ate^

— {Ai = Atm^, M = Atei)
We start with the combination tjj = lUf,ip = v^,A = Ate^, ^ = ^TMl'-

(1)

(2)

Note that

d?xNo

^oji (Por) (a-x

^1/2 (^)

t

lol--. Ji{i^TEr)Y^i[x)

ILOTE

a.x^,f{x) = -^,rj

M

with /' = / — A = J— ^A and furthermore

707

1 \ / a
-1 \-a

a
a

14

which yields

Nte

iuJTE

lUJTE

-nojiipor)iXjiiParW{!/,^a ■ y^^f ] h{ujTEr) (67)

Consider the relation

^Mit^^M, _ . ■^.j,+z.,+M / 3(2Ji + 1)(2J2 + l)(2Li + 1)(2L2 + 1)

27r

{Li Ji i
L J 1

{ Ji-MiJ2M2\JM )Yj\, (68)
and furthermore the orthogonality

j Ylf^Tix) ■ Yl§{x)dn = Sj,j6l'lSm'm (69)

and also

Yi§*{x) = i-iy+^+'^^^'YZj'\x). (70)
These relations allow us to express

' ■Y{{^*dn (71)

^01/2 "^-^i'J

It is now obvious that only the term J = 1, L = 1 of the sum in eq. (j68p survives. Furthermore,
( L1OL2OILO ) = ( 00r0|10 ) tells us that /' = 1, which allows only the modes J = ^, A = -1
and J = |, A = 1. The factor ( Ji - M1J2M2I JM ) = ( i - iJM|l - 1 > requires M = -i.
The expression

■Y-^'*dn (72)

requires I = 0,2, which allows the modes J=i, A = — 1, J=|, A = l and J = |, A = — 1.
The last mode, however, is prohibited by the second Clebsch-Gordan coefficient in eq. (j68p .
Thus, the dOa^-integral constraints the values for J, A, M in eq. (l62p . while the sum goes over all
possible n.

(1) of eq. (j62p takes the form

J = 1/2, A = -1 :

jiiujTEr) [-iio(^o?')f^Qji(war)Q(0, 1/2, V2 II 1,1/2,-1/2 II 1,1)
+noji{ujor)jo{oJar)a{l, 1/2, 1/2 II 0, 1/2, -V2 II 1, 1)] (73)

15

J = 3/2, A = 1 :

jiiujTEr) [-iio(^o?')ji(war)Oaa(0, 1/2, V2 II 1,^2, -V2 II 1,1)
-if]oJi(wor)j2(a;„r)a(l, 1/2, 1/2 || 2,3/2, -I/2 || 1, 1)] (74)

where we left out the expression NoNa j drr"^ for convenience.
We introduce

a{L,,J,,M^ II L2, J2,M2 II L,J) = (^-ly+M^+Ah

3(2J, + 1)(2J2 + 1)(2L, + 1)(2L2 + 1) ^ ^^^^^^^^^ ^

2tt

Ll Jl i I

^2 J2 I > ( ^1 - Ml J2M2I JM > (75)

L J 1

A hst of expressions for a can be found in Table [1]
(2) of eq. (j62p becomes:

J = 1/2, A = -1 :
[2

^JoioJTAir') [ijo(war')jo(^or')a(0, 1/2,-1/2 || 0,1/2,1/2 II 0,1)
-J^„ji(a;„r')iJ^oJi(^^r')a(l,V2,-i/2 || 1,1/2,1/2 II 0,1)]

^j2{u:TMr') [-Qaji{uJar')i0.oji{uJor')

0(1,1/2,-1/2 II 1,1/2,1/2 II 2,1)] (76)

(77)

J = 3/2, A = 1
0(1,3/2,-1/2 II 1,1/2,1/2 II 0,1)]

- y|j2(™r') [-fj2(war')io(^^or')a(2,3/2,-i/2 || 0,1/2,1/2 || 2,1)

-0„ji(a;„r')ifloii(wor')a(l,3/2, -I/2 || 1, 1/2, 1/2 || 2, 1)] (78)

where we left out the expression -NoNa^^ J dr'r''^. The additional minus sign comes from
the fact that Y^f* = - YqI^'^ and Y^f* = -Y^^^" , while y^f * = Y{{''^ .

In order for the combinations in (|6ip to give a nonzero contribution, the spins have to be
aligned as u^A^A^v^ or uiA^A^vi. This corresponds to Mi — Mi,M2 — >• — M2 in a ([75]) .
Because of

a(Li, Ji,Mi II L2, J2,M2 II L, J) _ ^_^Yi+J2-J+i

a{Li, Jl, -Ml II L2, J2, -M2 II L, J)

where Mi, M2 G {-I/2, 1/2} (79)

16

the signs from the two integrals cancel. Thus, the combinations with inverse magnetic quantum
number give the same contribution.
Next, we consider the expression

■U07 • AxEUaUaJ ' AtmVq

(80)
(81)

where we inserted the definition for the charge-conjugated particle state. Since this expression
has the form of a c-number, we transpose it by Hermitian conjugation and complex conjugation.
Hermitian conjugation gives

Uo\-l^)i-i)i-7-A*TMh'^Uc,uii-j-A*TEh°uo.
Complex conjugation gives

(82)

ul'j'^i{j ■ Atm

A

TM

i*Jij-A

TE

Ate

2\ *
7 )7

Ate

7 )7 %

(83)
(84)
(85)

(87)

= - u,Ji(7 • i*rAf)7^7°^^Q'"a 7^7^(7 • ^TE -

= - uIj°{j ■ ATM)i7'^U*aU*J-f'^{'J ■ ATEh'^l^ul

= - Uo{j ■ ATM)VaU*J-f^^'^{j ■ ATEh^UQ
= Uo{^ ■ ATM)VaU*J-r'^J°i{^ ■ ATEh^ul
= Mo (7 • ATM)VaVa {l ' Ate)vo .

For the last step, we have used v = (27^^*)^ 7*^ = iu*^'y'^j^.

Taking into account the definition of the confined propagator (|39|) as well as the time integration
with the recoil correction, one finds that the combinations with TE -f-)- TM give the same
contribution. Thus, there is an overall symmetry factor of 4 for the different combinations of
wavefunctions.

uoA-teSf4-tmVo, particle propagation:

J = 1/2, A

-1

-i^ay -^joii^or)jiiuJar)ji{uJTEr) - ir^oy -^jii^^or)joiuJar)ji{u}TEr)j

O.a0.oji (worOil {i^ar')j2 i^^TAir']

(89)

17

J = 3/2, A = 1 :

\/2

+i ^^^ ^a^oji {uJor')ji (War')j2 i^^TAir')

(90)

uq4^xeSfAtmVq, antiparticle propagation:

J = 1/2, A = 1 :

(91)

J = 3/2, A = -1 :

i\l i^jQ{^ar)ji{^ar)ji{iOTEr) + ^J]o^^aii(^o?')j2(war)ii(cjTEr)

3 V (^o^-Oii (wQr')io (o^TMr-')

a/2

-i^o 12^ ^^ (^o^^Oii {^cxr')j2 (uJTMr')

\ \ ~^^°4^-?o('^o''')-?2(war')j2(wTJ\/r')

(92)

For convenience, we have left out the expression —i NqN"^ . — / drr"^ / dr'r'"^. Also,

' iujte^^tm

one has a factor of 2(7^ ' ^ ' ^ ' y 1' '^^ich arises from eq. (|12p . the symmetry in the combinations,
the wavefunction symmetrization and the color matrix trace, respectively. For the confined
propagator, it is important to note the J, A quantum numbers, because the modes depend
on these quantum numbers as well as on the n quantum number, which is summed over.

18

T

T

A/f

T

-^2

i\/r

1

1/2

1/2

1/2

-1/2

n

1/9

1/9

1

X

3/9

-1/9

1

1/2

1/2

2

3/2

-1/2

1/2

-1/2

1/2

1/2

1

3/2

-1/2

1

1/2

1/2

1

1/2

-1/2

1

1/2

1/2

1

3/2

-1/2

1

1/2

1/2

1/2

1/2

2

3/2

-1/2

1

1/2

-1/2

1

1/2

1/2

1 1

1 1

1 1

1

1

1

1

Table 1: a coefficients

a

V2
120F

2

3v^
1

V2

V2
60F

References

[1
[2:

[3]
[4]
[5]

[6]
[7]
[8]
[9]
[10]

[11]
[12]

[13]

H. Fritzsch, M. Gell-Mann and H. Leutwyler, Phys. Lett. B47 (1973) 365.
D. J. Gross and F. Wilczek, Phys. Rev. Lett. 30 (1973) 1343.
H. D. Politzer, Phys. Rev. Lett. 30 (1973) 1346.
K. G. Wilson, Phys. Rev. D 10 (1974) 2445.

V. Mathieu, N. Koclielev and V. Vento, Int. J. Mod. Phys. E 18 (2009) 1 [arXiv: 0810.44531
[hep-ph]].

V. Crede and C. A. Mever. [i;^v:0812.0600l [hep-ex].

D. L. Scharre et al, Phys. Lett. B 97 (1980) 329.

H. Fritzsch and P. Minkowski, Nuov. Cim. 30A (1975) 393.

R. L. Jaffe and K. Johnson, Phys. Lett. B 60, 201 (1976).

S. Okubo, Phys. Lett. 5, 1975 (1963); G. Zweig, in Development in the Quark Theory of
Hadrons, edited by D.B. Lichtenberg and S.P. Rosen (Hadronic Press, Massachusetts, 1980);
J. lizuka. Prog. Theor. Phys. Suppl. 37, 38 (1966).

O. V. Maxweh and V. Vento, Nucl. Phys. A 407 (1983) 366.

A. Messiah, Mecanique Quantique, vol. 2, (Dunod, Paris 1962).

T. A. DeGrand, R. L. Jaffe, K. Johnson and J. E. Kiskis, Phys. Rev. D 12 (1975) 2060.

19

C. J. Morningstar and M. J. Peardon, Phys. Rev. D 60, 034509 (1999)
|arXiv:hep-lat/9901004] .

H. B. Meyer and M. J. Teper, Phys. Lett. B 605 (2005) 344 |arXiv:hep-ph/0409183

H. B. Meyer, arXiv:hep-lat/0508002!

Y. Chen et al, Phys. Rev. D 73 (2006) 014516 |arXiv:hep-lat/0510074 |.

J. Sexton, A. Vaccarino and D. Weingarten, Phys. Rev. Lett. 75 (1995) 4563
I a rXiv:hep-lat /951 0022| .

V. Mathieu and V. Vento, larXiv:0910.0212l [hep-ph].

H. Y. Cheng, H. n. Li and K. F. Liu, Phys. Rev. D 79 (2009) 014024 [arXiv: 081 1.25771
[hep-ph]].

J. Kuti, Nucl. Phys. Proc. Suppl. 73 (1999) 72 fa rXiv:hep-lat /981 1021| .
C. E. Carlson and T. H. Hansson, Nucl. Phys. B 199 (1982) 441.

R. Escribano, Eur. Phys. J. C 65 (2010) 467 |arXiv:0807.4201l [heo-ph]].

D. Binosi and L. Theussl, Comput. Phys. Commun. 161 (2004) 76 |a rXiv:hep-ph/0309015 |.

D. A. Varshalovich, A.N. Moskalev, V. K. Khersonskii, Quantum Theory of Angular Mo-
mentum, (World Scientific, Singapore 1988).

20